Small-Scale Pilot Testing of Stormwater Treatment Systems to Meet Numerical Effluent Limits in the Lake Tahoe Basin

Transcription

1 California State University, Sacramento (CSUS) University of California, Davis (UCD) California Department of Transportation (Caltrans) Small-Scale Pilot Testing of Stormwater Treatment Systems to Meet Numerical Effluent Limits in the Lake Tahoe Basin Reprinted with permission from: Proceedings of WEFTEC.5, the 78th Annual Water Environment Federation Technical Exhibition and Conference (CD-ROM), Washington, DC, Oct 3-Nov 2, 25. Copyright 25 Water Environment Federation: Alexandria, Virginia. Authors: Jeffrey Hauser, ECO:LOGIC Engineering, Rocklin, California Jeffry Curtis, ECO:LOGIC Engineering, Rocklin, California John Johnston, Department of Civil Engineering, California State University, Sacramento Dipen Patel, Office of Water Programs, California State University, Sacramento Mark Keisler, California Department of Transportation, Sacramento Disclaimer: This work reflects the author s opinions and does not represent official policy or endorsement by the California Department of Transportation, the California State University, or the University of California. Storm Water Program CSUS Office of Water Programs 6 J Street, Modoc Hall, Room, Sacramento, CA PP63

2 SMALL-SCALE PILOT TESTING OF STORMWATER TREATMENT SYSTEMS TO MEET NUMERICAL EFFLUENT LIMITS IN THE LAKE TAHOE BASIN Jeffrey Hauser *, Jeffry Curtis, John Johnston 2, Dipen Patel 3, Mark Keisler 4 ECO:LOGIC Engineering, 3875 Atherton Road, Rocklin, CA Department of Civil Engineering, California State University Sacramento, CA 3 Office of Water Programs, California State University Sacramento, CA 4 California Department of Transportation, Sacramento, CA ABSTRACT To stem the continuing loss of clarity in Lake Tahoe, stringent numerical discharge requirements for storm water discharges in the Tahoe basin have been promulgated. In 28, discharges to surface water must meet limits for turbidity (2 NTU), total phosphorus (. mg/l), total nitrogen (.5 mg/l), total iron (.5 mg/l), and oil and grease (2 mg/l). The California Department of Transportation (Caltrans) has undertaken a multi-year small-scale storm water treatment pilot project to identify, test, and evaluate potential treatment technologies to meet the effluent limits. Treatment technologies investigated include simple non-mechanized sedimentation and granular media filtration systems (with and without chemicals), plus more complex mechanized systems involving combinations of coagulation, flocculation, sedimentation, filtration, and ion exchange. During the first two years of operation covered in this paper, thirty different non-mechanized systems and eight mechanized systems were tested. In terms of turbidity and phosphorus removals, conventional storm water treatment systems (detention and filtration without chemicals) were consistently unable to meet the discharge limits, whereas similar non-mechanized systems preceded by chemical coagulation performed much better, frequently meeting the limits. Sedimentation without chemicals followed by filtration through activated alumina or expanded shale adsorptive media also performed well in removing turbidity and phosphorus, but resulted in elevated ph and dissolved aluminum concentrations. Most of the mechanized systems were successful in meeting the turbidity and phosphorus limits, though these systems are not well suited to the roadside environment. The nitrogen limit was not consistently met by any of the systems tested. KEYWORDS Stormwater treatment, Lake Tahoe, turbidity, phosphorus, stormwater BMP, chemical treatment, coagulation, filtration, sedimentation INTRODUCTION Lake Tahoe is known worldwide for its stunning clarity and deep blue water. Over the past four decades, though, the clarity of the lake has declined, partially due to algae growth, and partially due to small particles being washed into the lake. In response, the Lahontan Regional Water Quality Control Board has prescribed stringent effluent limits for all storm water discharges (Table ), which are scheduled to take effect in 28. These limits pose a significant challenge 35

3 to agencies such as the California Department of Transportation (Caltrans), which owns and maintains over 68 miles of roadway in the basin. As seen in Table, constituent concentrations in highway runoff exceed, sometimes greatly, the legal effluent limits. In 2, Caltrans initiated a multi-year small-scale pilot testing program at the lake. The objective of this program is to identify, test, and evaluate potential treatment methods for highway storm water runoff to meet the effluent limits for surface water discharge. Surface water effluent limits are the primary target because Caltrans has limited access to land for storm water infiltration. In this paper, an overview of program activities and results during the 2-2 and 22-3 wet seasons is presented. These are referred to as Year and Year 2 respectively in this paper. Details can be found in reports available from Caltrans (Caltrans 23b, 23c). Table - Highway Runoff Characteristics and Legal Effluent Limits in the Tahoe Basin Parameter Tahoe Basin Highway Runoff (Mean Values) (a) Effluent Limits for Storm Water (b) Discharge to Discharge to Land Surface Water (Infiltration) Turbidity (NTU) Total Phosphorus (mg/l as P) 2.. Total Nitrogen (mg/l as N) Total Iron (mg/l as Fe) Oil and Grease (mg/l) (a) Caltrans, 23a (b) LRWCB, 994 PILOT TREATMENT FACILITIES OVERVIEW In the Fall of 2, Caltrans constructed a 24-ft 2 small-scale pilot storm water treatment facility at the South Lake Tahoe Maintenance Station (Figure ). This facility houses two categories of treatment systems: ) non-mechanized systems requiring little or no power or operator attention that could be deployed in a roadside setting, and 2) mechanized systems utilizing powered mechanical equipment requiring operator attention and probably a building enclosure. In general, the non-mechanized systems involve relatively slow-rate treatment methods that require a large footprint area per unit of flow to be treated and would be mostly applicable to rural settings where land is readily available. Mechanized systems, in contrast, provide relatively high-rate treatment and would be more appropriate for urban settings where large volumes of flow must be treated in a small area. All of the non-mechanized experimental treatment systems were based on sedimentation and/or granular media filtration with and without chemical addition. A total of 3 different treatment systems were tested, as listed in Table 2. Sedimentation tanks were operated in batch mode for convenience. Filters were operated as slow rate systems without backwash, as they would be in 352

4 a roadside installation. Key data for the various media are shown in Table 3. The different sandbased media were expected to provide treatment by mostly physical means (straining, impaction, etc.), whereas the remaining media were expected to adsorb ammonia nitrogen or phosphorus. All of the mechanized treatment systems included one of two configurations of chemical coagulation, flocculation, and sedimentation as the first or only treatment step: ) a batch version of an Actiflo (Kruger, Inc., Cary, NC) process including the use of ballast sand, and 2) conventional operation without ballast sand. Subsequent treatment was investigated in a stepwise manner, with the second step involving either a pressure sand filter or a proprietary high-rate synthetic media filter (Fuzzy Filter, Schreiber Wastewater Treatment Technologies, Trussville, Al). The third treatment step was ion exchange using separate cation and anion resin beds. In total, eight treatment systems were investigated as delineated in Table 4. Figure Small-Scale Pilot Treatment Facility Showing Exterior with Storm Water Storage Tanks (Left) and Interior with Non-Mechanized Treatment Units (Right) The chemicals used in the various treatment systems and the doses and methods of application are discussed in following sections. The pilot facilities were operated in response to runoff events (rainfall or snowmelt). When runoff occurred, a water truck was dispatched to collect storm water from one or more catchment basins in the Lake Tahoe area. The collected storm water was trucked to the pilot treatment facility and stored in a 25, L (6,5 gal) polyethylene storage tank until processed through the treatment units. While in storage, the storm water was continuously mixed with a submersible mixer. NON-MECHANIZED TREATMENT SYSTEMS CONFIGURATION AND OPERATION The sedimentation units and gravity filters were constructed from 2.-m (7-ft) tall,.76-m (3-inch) diameter polyethylene tanks on steel stands (Figure 2). These were followed by.-m (3.5-ft) tall,.76-m (3-inch) diameter polyethylene effluent collection tanks at floor level. 353

5 For each batch run, the sedimentation tanks were filled to a depth of.8 m (6 ft) and then allowed to settle for either 2 or 24 hours (Table 2). The sedimentation tank effluent sampling port and outlet were located.9 m (3 ft) above the tank bottom. For those treatment systems including filtration, only the upper.9 m (3 ft) of settled storm water in the settling tank was transferred to the filter. 354

6 Table 2 Non-Mechanized Treatment Systems No. Chemical Sed. Time, Hrs Filter Medium Filter Loading Rate Filter Hydraulic Condition Year (-2) Year 2 (2-3) Sedimentation Sedimentation with Chemicals 3 Chitosan PAC Filtration Fine Sand Fast Free-Drain Coarse Sand Fast Free-Drain Zeolite Fast Free-Drain Activated Alumina Fast Free-Drain Aluminum Oxide Fast Free-Drain Filtration with Chemicals PAM --- Fine Sand Fast Free-Drain PAM --- Coarse Sand Fast Free-Drain 2 PAC --- Fine Sand Fast Free-Drain 3 PAC --- Coarse Sand Fast Free-Drain Sedimentation and Filtration Conc. Sand Fast Free-Drain Fine Sand Fast Free-Drain Fine Sand Slow Free-Drain Fine Sand Slow Submerged Fine Sand Fast Submerged Fine Sand Slow Free Drain Fine Sand Slow Submerged Expanded Shale Slow Submerged Limestone Slow Submerged Wollastonite Slow Submerged Activated Alumina Slow Submerged Sedimentation and Filtration with Chemicals 25 PAC 2 Fine Sand Fast Free-Drain 26 PAC 2 Fine Sand Slow Free-Drain 27 PAC 2 Coarse Sand Fast Free-Drain 28 PAC 2 Zeolite Fast Free-Drain 29 PAC 2 Activated Alumina Fast Free-Drain 3 PAC 2 Aluminum Oxide Fast Free-Drain 355

7 Table 3 - Filter Media Tested Filter Media Characteristics / Description Effective Size (a) (D, mm) Uniformity Coefficient (a) (D 6 /D ) Fine Sand mm.47.5 Coarse Sand.8- mm.7.9 Concrete Sand ASTM C Activated Alumina Alcoa DD-2 28 x Aluminum Oxide White Aluminum Oxide, 3 Grit.45.4 Zeolite Clinoptilolite.45.4 Limestone Limestone #4 Sand Expanded Shale Utelite Fines Wollastonite (a) Before conditioning Wollastonite Tailings (media passing #5 sieve removed) Table 4 - Mechanized Treatment Systems No. Configuration Year Year 2 Actiflo 2 Actiflo + Fuzzy Filter 3 Actiflo + Pressure Sand Filter 4 Actiflo + Fuzzy Filter + Ion Exchange 5 Actiflo + Pressure Sand Filter + Ion Exchange 6 Conventional Coagulation, Flocculation and Sedimentation 7 Conventional Coag/Floc/Sed + Pressure Sand Filter 8 Conventional Coag/Floc/Sed + Press. Sand Filter + Ion Exch. 356

8 Figure 2 Typical Non-Mechanized Treatment System Chemical Dosing in the Rear Granular Media Filter Settling Tank Effluent Collection Tank In Year, each experimental run included two 46 L ( gal) batches. After the first batch was settled, the settling tanks were drained down to the.9 m (3 ft) depth by discharging to the filters (for those systems including filtration). The tanks were then refilled to the.8 m (6 ft) depth with fresh storm water, allowed to settle, and discharged again. For systems without filters, the sedimentation tanks were operated in the same manner, but discharged to drain. In Year 2, the sedimentation tanks were filled and discharged only once per run, meaning that each run consisted of only one 46 L ( gal) batch. The filtration units contained 6 mm (24 inches) of filter media above 23 mm (8 inches) of pea gravel, separated by a layer of geotextile fabric. A perforated pipe extending across the unit within the pea gravel was used to collect and discharge filter effluent. After placement in the filters, the media were rinsed to remove fines by thoroughly flushing with tap water using a section of 9-mm (¾-in) PVC pipe connected to a hose. The rinsing pipe was moved up and down through the media bed at various locations, dislodging media fines and dirt particles. Rinsing was continued until the turbidity of water draining through the media was 2 NTU or less. During the first year of operations, the transfer of water from the settling tank to the filters was accomplished by opening a 38-mm (.5-inch) sedimentation outlet valve and allowing the water to drain down through the filter as fast as allowed by filter permeability. Filters not preceded by settling were loaded by pumping storm water directly to the filters at a rate of 9 L/min (24 gpm), which resulted in a filter surface loading rate of 2 L/min m 2 (4.9 gpm/ft 2 ). Both cases are designated as fast loading in Table 2. Also in Year, all of the filter effluents were discharged freely to the atmosphere from the underdrain outlet pipe at the underdrain elevation, resulting in a free-draining filter hydraulic condition (Table 2). In the second year, all of the filters followed sedimentation and most were loaded at a slow rate by using a peristaltic pump to transfer the 46 L ( gal) batch feed volume from the settling tank over a period of six hours at a rate of.6 L/min (.36 gpm), resulting in a surface loading rate of 2.5 L/min m 2 (.62 gpm/ft 2 ). Some of the filters were operated in the free- 357

9 draining hydraulic condition, while others were operated in a submerged condition (Table 2), meaning that the filter effluent pipe was extended upward to a level just above ( mm +/-) the filter media surface before discharge to atmosphere. This assured that the media bed was always saturated, which was intended to cause the influent to be distributed uniformly over the entire filter area. As indicated in Table 2, three different chemicals were used in various non-mechanized treatment systems, including a polyaluminum chloride (PAC), a polyacrylamide (PAM), and chitosan. The PAC product used was PASS-C (Eaglebrook Inc., Matteson, IL), which was chosen based on preliminary jar testing with various storm water samples prior to the first pilot run. Several other PAC products are also known to provide effective storm water treatment. PAC was applied by using a peristaltic pump to meter the chemical into an in-line static mixer in the treatment unit influent piping. This provided uniform dosing at the desired rate. During Year, PAC was applied at a constant dose of mg/l (as liquid product), based on the results of preproject jar testing. In Year 2, the PAC dose was based on jar testing conducted for each run (i.e., optimized for that particular storm water) and ranged from 75 to 2 mg/l. A granular anionic PAM, Superfloc A-836 (Cytec Chemical, West Paterson, NJ), was used in two of the non-mechanized treatment systems during Year. A passive tea bag dosing system was employed. This involved suspending a small (4 mm x 5 mm) geotextile fabric bag containing the chemical in the influent flow stream. Chitosan, which was used in one non-mechanized system during Year 2, is a refined natural polymer extracted from chitin, a substance found in the shells of crustaceans such as crabs, shrimp, and lobsters. The chitosan used was obtained from Natural Site Solutions, Redmond, WA. Both solid (Gel-Floc TM ) and liquid (Liqui-Floc TM ) products were used as discussed later in this paper. The solid form was used in the first two runs and was applied using the same tea bag approach as used for PAM. For the subsequent four runs, the liquid product was used at a constant dose rate of.5 or. mg/l (dry basis) using a metering pump and static mixer. MECHANIZED TREATMENT SYSTEMS CONFIGURATION AND OPERATION The equipment used in the mechanized treatment systems is illustrated in Figure 3. Coagulation, flocculation, and sedimentation (Actiflo and conventional) for the mechanized treatment systems were accomplished in a 95-mm (36-inch) square, 37-mm (54-inch) tall fabricated stainless steel tank. The tank was fitted with a vertical-shaft mixer that was controlled to attain the desired speed and mixing intensities using a variable frequency drive. For each run, the tank was filled with 984 L (26 gal) of storm water and then operated through a timed sequence of micro-sand (Actiflo only) and chemical additions, rapid and slow mixing, and then settling as indicated in Table

10 Figure 3 Mechanized Treatment Systems Table 5 - Operational Steps for Mechanized Treatment Coagulation, Flocculation, and Sedimentation Actiflo Step Activity Mixer Speed, rpm Duration, Min Conventional Mixer Speed, Rpm Duration, min Add Ballast Sand (a) Add PAC Rapid Mix Add Polymer Solution Rapid Mix Slow Mix Settle 3 to 3 (a) Actiflo only; 5. L of sand were added in each run. Two chemicals were used to aid coagulation and flocculation. Coagulation was accomplished using the same PAC as previously described. The PAC dose during Year was mg/l (as liquid product) in the first three runs and 25 mg/l in the next three runs. Flocculation was promoted by the use of an anionic polymer, Magnafloc LT25 (Ciba Chemicals, Suffolk, VA). The polymer dose during Year was.8 mg/l (dry weight basis) in the first three runs and 359

11 . mg/l in the next three runs. In Year 2, appropriate dose rates for both the PAC and the polymer were determined by jar testing for each run. Doses ranged from 75 to 2 mg/l for the PAC and.5 to.5 for the polymer. After settling, the upper portion of the settling tank was pumped at a constant flow rate of 9.8 L/min (2.6 gpm) through either the Fuzzy Filter or the pressure sand filter, and then through the ion exchange columns. In Year, the Actiflo process was used with both the Fuzzy Filter and the pressure sand filter. In Year 2, the Actiflo process was used only with the Fuzzy Filter and the conventional coagulation, flocculation, and sedimentation process was used only with the pressure sand filter. In all cases, a total of 38 L ( gal) of settled storm water was pumped through each filter for each run. The Fuzzy Filter unit was constructed of 2-mm (4-inch) diameter clear PVC pipe, approximately 94 mm (36 in) long with capped (blind flange) ends. Inside that vessel, the filter bed was comprised of a 762 mm (3 in) depth of wet (water saturated) fiber spheres compacted to 533 mm (2 in) using perforated plastic compression plates. Influent piping was connected to the bottom of the unit and effluent piping to the top to provide an upflow configuration. The surface loading rate for this unit was 2 L/min m 2 (3 gpm/ft 2 ). The pressure sand filter was a prefabricated steel unit manufactured by Process Efficiency Products, Mooresville, NC (Model CL-2). The filter contained a filter sand (.55 mm effective size) bed with a surface area of.93 m 2 (. ft 2 ) and a depth of 35 mm (2 in). The surface loading rate for this filter was 5 L/min m 2 (2.6 gpm/ft 2 ). The cation and anion exchange columns and housings were obtained from American Filter Works, Los Angeles, CA. The volume of ion exchange resin in each column was,9 cm 3 (4 in 3 ). The influent flow rate of 9.8 L/min (2.6 gpm) represented a volumetric loading rate equal to 5.2 bed volumes per minute. SAMPLING AND LABORATORY ANALYSIS Physical and chemical data were collected during each experimental run to assess treatment efficiencies. The water quality parameters monitored during Year 2 are shown in Table 6. In Year, additional parameters, including minerals (calcium, potassium, sulfate, chloride, magnesium, sodium, boron, and silica) and metals (total and dissolved cadmium, chromium, copper, lead, nickel, and zinc) were monitored. 36

12 Table 6 Water Quality Parameters Monitored in Year 2 Field-Measured Parameters Laboratory-Measured Parameters Specific Conductance Alkalinity Total Total Phosphorus (Filtered) ph Total Dissolved Solids Total Phosphorus (Un-Filtered) Turbidity Total Suspended Solids Ortho-Phosphate (Filtered) Temperature Nitrate Nitrogen Ortho-Phosphate (Un-Filtered) Nitrite Nitrogen Ammonia Nitrogen Total Kjeldahl Nitrogen (Filtered) Total Kjeldahl Nitrogen (Un-Filtered) Total Organic Carbon Oil and Grease Total Aluminum (Unfiltered) Total Aluminum (Filtered) Acid Soluble Aluminum Total Iron (Unfiltered) Total Iron (Filtered) In general, samples of the influent and the effluents of the various treatment units were composites formed from multiple grab samples collected manually over the course of an experimental run. For the final step of the non-mechanized treatment systems (the filter effluent for the systems with sedimentation and filtration), the entire effluent volume for each half-run in Year and for each run in Year 2 was collected in a sample collection tank as previously described. A grab sample was then taken from the filled tank (while being stirred) at the completion of the half run or run. In Year, the grab samples from each half run were composited to characterize the effluent for the entire run. Complete descriptions of the sampling procedures followed during Year and Year 2 can be found in the respective Monitoring and Operations Plans (Caltrans, 22 and 23d). Field sample processing activities included splitting the sample into multiple sample containers for various contaminant analyses and filtering various samples for dissolved analyses. All sampling and laboratory analyses were in accordance with the sampling and analytical requirements established by Caltrans for the monitoring of storm water (Caltrans, 2). Environmental contamination of the samples during processing was minimized by making use of clean sampling techniques (Caltrans, 2). RESULTS OVERVIEW The small-scale pilot treatment project discussed in this paper included assessments of the hydraulic characteristics and pollutant removal performances of the various treatment systems. All systems were analyzed with regard to all of the parameters listed in Table 6 for Year 2 and 36

13 the additional parameters previously discussed for Year. In the space available here, it is impossible to discuss all of the results. Therefore, the focus of this paper is on turbidity and total phosphorus because these are the most important constituents for Lake Tahoe clarity. Algae and small particles washed into the lake are thought to contribute directly to loss of clarity, and Lake Tahoe is considered to be phosphorus-limited with regard to algae growth. In the sections that follow, the hydraulic performances of the various filter systems are discussed, turbidity and phosphorus results for all of the treatment systems are presented and discussed in detail, followed by consideration of other constituents. Complete information on parameters monitored but not considered in detail here can be found in Caltrans 23b and 23c. Six experimental runs were completed in each of the first two years of pilot operations. For various reasons, however, some of the treatment units were not operated in some of the runs. Therefore, the graphs in which treatment results are shown in the following sections have different numbers of data points for the various treatment units. HYDRAULIC PERFORMANCE OF FILTER SYSTEMS As previously noted, in Year, a total of.8 m (6 ft) of storm water was applied during each run to each of the non-mechanized filters. In Year 2, only.9 m (3 ft) was applied during each run. While the filters were in operation, filter head loss and/or drain-down times were observed and recorded. In Year, all of the filters were operated in the fast-load, free-drain configuration and many were not preceded by sedimentation basins. In Year 2, different loading rates and hydraulic conditions were tested, but all of the filters were preceded by sedimentation basins. None of the non-mechanized filters following sedimentation failed hydraulically. Many of the filters not preceded by sedimentation, however, failed because of media blinding or clogging. Failure was judged to occur when the applied water remained on top of the media several hours after application. Except as otherwise noted later in this paper, after a filter failed, it was not used in subsequent runs. The filters that failed are listed in Table 7, together with the time of failure (Run Number, Fill Number) and the depth of storm water and solids load that had been applied from the beginning of the study to the point of failure. The non-mechanized filters that were tested without chemical addition failed after 6.4 to 8.2 m (2 to 27 ft) of water and 2. to 2.3 kg/m 2 (.4 to.47 lb/ft 2 ) were applied. This is of concern since it is anticipated that the average annual hydraulic loading for filters in the Lake Tahoe area will be in the range of 27 to 9 m (9 to 3 ft), depending on the specific configuration used. It is desirable that non-mechanized filters be able to operate for at least one full year without maintenance. As shown in Table 7, the filters that received chemicals failed even faster, most likely due to flocs accumulating on the filter surface. The only filters not preceded by sedimentation that did not fail were those using coarse sand as a filter media. 362

14 Table 7 Non-Mechanized Filter (Without Sedimentation) Hydraulic Failures (Year ) Filter Medium Chemical FailureRun, Fill Applied Water, m (ft) Applied TSS (a), kg/m 2 (lb/ft 2 ) Fine Sand None 5, 7.3 (24) 2.2 (.44) Zeolite None 5, (27) 2.3 (.47) Activated Alumina None 4, (2) 2. (.4) Aluminum Oxide None 5, (27) 2.3 (.47) Fine Sand PAM,.9 (3).34 (.7) Coarse Sand PAM 2, 2.7 (9).72 (.5) Fine Sand PAC 4, (2) 2. (.4) (a) TSS = total suspended solids, expressed as mass per unit of filter surface area. Since the storm water influent to the mechanized filters was always pre-treated by coagulation, flocculation, and sedimentation and had low turbidity, no significant head loss buildup was noted in either the Fuzzy Filter or the pressure sand filter. Therefore, the pressure sand filter was not backwashed in Year or in Year 2. The Fuzzy Filter media, which were visible through the clear PVC vessel wall, became discolored (gray/brown) and were replaced with new media before Run 4 of Year. As a subsequent standard procedure, the Fuzzy Filter media were replaced with cleaned or new media before Run 6 of Year and before all runs in Year 2. TURBIDITY REMOVAL As noted in Table, the average turbidity of Tahoe area highway runoff is about 5 NTU, while the effluent limit for discharge to surface waters is 2 NTU. In the paragraphs that follow, the turbidity removal results for the various treatment systems are presented. Throughout the presentation, the performance of each treatment unit is illustrated in graphical form, with the influent concentration plotted on the horizontal axis and the effluent concentration plotted on the vertical axis (see Figure 4). A diagonal line indicates equal influent and effluent values. Data points falling below the diagonal line indicate that constituent removal (i.e., treatment) occurred. Also on each graph is a dashed horizontal line that indicates the regulatory requirement for discharge to surface waters (2 NTU). Non-Mechanized Sedimentation With and Without Chemical Assistance Turbidity removal performances of the non-mechanized sedimentation systems tested in Year 2 are shown in Figure 4. The systems represented by Figures 4a and 4e were tested in Year also, giving similar results to those shown. 363

15 Figure 4 Turbidity Results for Non-Mechanized Sedimentation Systems a. 2 hr Sedimentation, Rapid Empty (Year 2) b. 2 hr Sedimentation, 6 hr Empty (Year 2) c. 24 hr Sedimentation, 6 hr Empty (Year 2) d. 24 hr Sedimentation with Chitosan (Year 2) Dose = mg/l Dose =.5 mg/l Dose = Passive e. 2 hr Sed. With Optimized PAC Dosing (Year 2) As can be seen in Figures 4a, 4b, and 4c, sedimentation without chemicals was generally ineffective. Lengthening the detention time improved turbidity removals only minimally (compare Figure 4c to 4b). Comparing Figures 4b and 4c, it appears that most sedimentation occurs within a relatively few hours and that long holding times (i.e., 24 hours) provide little extra benefit. Providing chemical coagulation in front of sedimentation greatly increased turbidity removals as seen in Figures 4d and 4e. When the dose was controlled, as it was with PAC and the liquid chitosan (the.5 and mg/l doses in Figure 4d), the legal limit was achieved or almost achieved with sedimentation alone. The passive (tea bag) dosing method used in this study did not allow 364

16 adequate dose control and this resulted in less efficient turbidity removals. It would appear that the passive dosing system tried here did not deliver a sufficient quantity of chitosan. It s important to note that in these experimental runs, the slow-mixing step commonly used in drinking water treatment plants was omitted. As previously described, the coagulant was injected, mixed in-line, and delivered directly to the sedimentation facility. Non-Mechanized Filtration Without Sedimentation As previously noted, all of these filters, except those containing coarse sand, failed hydraulically before completing even five runs in Year. From a practical point of view, then, this is not a viable approach. Nevertheless, examination of the water quality results gives some useful insights. Turbidity results for the runs that were completed or partially completed are shown in Figure 5. The coarse sand filter without chemical addition (Figure 5b) didn t clog, but neither did it provide much treatment. Even with PAC, coarse sand was only moderately effective (Figure 5g). In contrast, the fine sand filter with chemical addition (Figure 5f) was very effective. Of the other filters not using chemicals (Figures 5a, 5c, 5d, and 5e), the activated alumina media (Figure 5d) showed the greatest promise. Non-Mechanized Filtration Following Sedimentation Without Chemicals Turbidity results for these systems are shown in Figure 6. All of the data are from Year 2, except for the concrete sand filter (Figure 6a), which was tested in Year. As noted earlier, none of these systems clogged during the study, suggesting that pre-filtration sedimentation is an essential component to successful operation, even when chemicals are not used. The systems represented in Figure 6 incorporate a number of different sedimentation and filtration conditions. Comparing turbidity results can give insight into how to optimize design and operations of these kinds of systems. Although a little difficult to see from Figures 6a and 6b, the concrete sand and fine sand filters performed about the same in the fast load, free drain mode with 2 hours of sedimentation. Performance improved when the filter loading was slowed and controlled (Figure 6c). Arranging the outlet to submerge the media bed (Figure 6d) further improved performance, though not by a large amount. The effects of increasing the sedimentation time from 2 to 24 hours can be seen by comparing Figures 6c and 6e, and Figures 6d and 6f. Increasing detention time had little or no effect, an observation consistent with what was seen in the sedimentation-only systems (Figures 4b and 4c). The results from the non-sand filters, expanded shale, limestone, Wollastonite, and activated alumina, are shown in Figures 6g through 6j. These were all slow-load, submerged filters following 24 hours of sedimentation. The equivalent sand filter is represented in Figure 6f. Under these conditions, the expanded shale (Figure 6g) and activated alumina (Figure 6j) media were more effective than sand and were the only media that consistently met or nearly met the 2 NTU regulatory limit, even without chemical addition. Limestone (Figure 6h) also performed better than sand, though not as well as expanded shale and activated alumina. Because of 365

17 difficulties in obtaining the Wollastonite media, it was available for only two runs (Figure 6i), and its performance was similar to that of fine sand (Figure 6f). Figure 5 Turbidity Results for Non-Mechanized Filtration Systems Without Sedimentation a. Fine Sand Filter (Year ) b. Coarse Sand Filter (Year ) c. Zeolite Filter (Year ) d. Activated Alumina Filter (Year ) e. Aluminum Oxide Filter (Year ) f. Fine Sand With Fixed PAC Dosing (Year ) g. Coarse Sand With Fixed PAC Dosing (Year )

18 Figure 6 (Part ) Turbidity Results for Non-Mechanized Filtration Following Sedimentation Without Chemicals a. Concrete Sand Filter, Fast Load, Free Drain Following 2 hr Sedimentation (Year ) b. Fine Sand Filter, Fast Load, Free Drain Following 2hr Sedimentation (Year 2) c. Fine Sand Filter, 6 hr Load, Free Drain Following 2hr Sedimentation (Year 2) d. Fine Sand Filter, 6 hr Load, Submerged Following 2 hr Sedimentation (Year 2) e. Fine Sand Filter, 6 hr Load, Free Drain Following 24 hr Sedimentation (Year 2) f. Fine Sand Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2)

19 Figure 6 (Part 2) Turbidity Results for Non-Mechanized Filtration Following Sedimentation Without Chemicals g. Expanded Shale Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) h. Limestone Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) i. Wollastonite Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) j. Activated Alumina Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) Non-Mechanized Filtration Following Sedimentation With Chemicals In Year, five different filter media following 2-hour sedimentation and coagulation using PAC were tested. All of the filters were fast-load, free-drain units which, as described above, was the least effective operational mode when chemical addition was not used. Despite this, all five systems consistently met or nearly met the 2 NTU turbidity limit (see Figures 7a, and 7c through 7f). In Year 2, the fine sand filter following 2-hour sedimentation and coagulation with PAC. was tested again. In this case, however, the free-drain filter was loaded over six hours and the PAC dose was optimized by running jar tests on each influent sample. As shown in Figure 7b, this system consistently met the 2 NTU turbidity limit. Although it is likely that the slow loading and optimized chemical dosing used in Year-2 enhanced the performance of the system represented in Figure 7b as compared to the system represented in Figure 7a, that conclusion can not be firmly made because different storm waters were tested (Year versus Year 2). Mechanized Treatment Systems Turbidity results for the mechanized coagulation, flocculation, and sedimentation processes tested in Years and 2 are shown in Figure 8. Recall that the mechanized systems incorporated both fast and slow mixing and use of both coagulant and flocculant chemicals. As can be seen in Figure 8, both the Actiflo and the conventional coagulation, flocculation, and sedimentation systems consistently met the 2 NTU turbidity limit, with the exception of one run in Year. Because downstream treatment processes added no meaningful improvement in water quality 368

20 except for the one run in Year, the results from these systems (listed as systems 2, 3, 4, 5, 7 and 8 in Table 4) are not shown here. In the one run in Year that the 2-NTU limit wasn t met after sedimentation, the influent turbidity of 84 NTU was reduced to 3 NTU in the Actiflo unit, and was further reduced to 25 and then 9 NTU in the subsequent Fuzzy Filter and ion exchange systems, respectively. Figure 7 Turbidity Results for Non-Mechanized Filtration Following Sedimentation With PAC a. Fine Sand Filter, Fast Load, Free Drain, Following 2 hr Sed. With Fixed PAC Dosing (Year ) b. Fine Sand Filter, 6 hr Load, Free Drain, Following 2 hr Sed. With Optimized PAC Dosing (Year 2) c. Coarse Sand Filter, Fast Load, Free Drain, Following 2 hr Sed. With Fixed PAC Dosing (Year ) d. Zeolite Filter, Fast Load, Free Drain Following 2 hr Sed. With Fixed PAC Dosing (Year ) e. Activated Alumina Filter, Fast Load, Free Drain Following 2 hr Sed. With Fixed PAC Dosing (Year ) f. Aluminum Oxide Filter, Fast Load, Free Drain Following 2 hr Sed. With Fixed PAC Dosing (Year )

21 Figure 8 Turbidity Results for Mechanized Coagulation, Flocculation, and Sedimentation a. Actiflo With Fixed Chemical Dosing (Year ) b. Actiflo With Optimized Chemical Dosing (Year 2) c. Conventional Coag., Floc. and Sed. With Optimized Chemical Dosing (Year 2) PHOSPHORUS REMOVAL As noted in Table, the average total phosphorus concentration in Lake Tahoe area highway runoff is 2. mg/l, while the effluent limit for discharge to surface waters is. mg/l. In the paragraphs that follow, the phosphorus removal results for the various types of treatment units are presented. Throughout the presentation, the performance of each treatment unit is illustrated in graphical form similar to the graphs used for the turbidity plots. In the following plots, however, dotted lines were added parallel to both the vertical and horizontal axes to indicate the analytical reporting limit (usually.3 mg/l). Laboratory results at or below the reporting limit are shown on the reporting limit line. In considering the experimental results, it is helpful to realize that the dissolved fraction of influent total phosphorus in Year almost always exceeded. mg/l, sometimes substantially, and was considerably lower in Year 2, never exceeding. mg/l. Non-Mechanized Sedimentation With and Without Chemical Assistance Total phosphorus treatment performances for the non-mechanized sedimentation systems tested in Years and 2 are shown in Figure 9. Sedimentation without chemical assistance is shown in Figures 9a through 9d. Sedimentation with chitosan is shown in Figure 9e and sedimentation with PAC is shown in Figures 9f (constant dose of mg/l) and 9g (dose optimized for each run). 37

22 As with the turbidity results, sedimentation without chemicals (Figures 9a through 9d) was ineffective at reducing total phosphorus. In contrast, all of the chemical-assisted systems produced substantial total phosphorus removals. Sedimentation after treatment with a mg/l dose of liquid chitosan (Figure 9e) and with PAC doses based on jar testing for the individual runs (Figure 9g) were effective at meeting the regulatory limit in almost all runs. Passive dosing of solid chitosan using the tea bag approach, liquid chitosan at a dose of.5 mg/l, and nonoptimized constant PAC dose of mg/l were less effective. Although proper dosing was most likely a factor contributing to the treatment results for the better-performing systems, since side-by-side testing of different dosing schemes on the same storm water was not completed, this conclusion can not be firmly made from the data presented. Non-Mechanized Filtration Without Sedimentation Total phosphorus results for the various fast-load, free-drain filters tested with and without chemical addition in Year are shown in Figure. As indicated in the figure, none of the filters were able to meet the regulatory total phosphorus limit, though the activated alumina filter (Figure d) and the fine sand filter with PAC (Figure f) provided somewhat better performance than the other units. These results generally mirror the turbidity results, except for the fine sand filter with PAC. In this case, turbidity removal (Figure 5f) was better than phosphorus removal, probably reflecting the fact that a substantial fraction of the influent phosphorus in Year was in dissolved form. Non-Mechanized Filtration Following Sedimentation Without Chemicals Total phosphorus treatment results for these systems are shown in Figure. All of the data shown are from Year 2, except for the concrete sand filter, which was tested in Year. As previously discussed, different sedimentation times and filter configurations were tested in Year 2, and these are noted on the graphs in Figure. Generally speaking, the effects of different sedimentation times and filtration conditions on phosphorus removal are similar to those on turbidity removal (Figure 6) and similar comments could be repeated here. These observations are consistent with the influent data showing that the majority of the influent phosphorus in Year and almost all of the phosphorus in Year 2 was in particulate form. The key observations to be made here are that the shale and activated alumina filters (Figures g and j) consistently met or nearly met the.-mg/l regulatory limit for total phosphorus. The limestone filter performance (Figure h) was similar, except for one run. Sand filters (without chemical addition) did not come close to meeting the total phosphorus limit regardless of how they were operated. 37

23 Figure 9 Total Phosphorus Results for Non-Mechanized Sedimentation Systems a. 2-hr Sedimentation, Rapid Empty (Year ) b. 2-hr Sedimentation, Rapid Empty (Year 2) c. 2-hr Sedimentation, 6hr Empty (Year 2) d. 24 hr Sedimentation, 6hr Empty (Year 2) e. 24 hr Sedimentation with Chitosan (Year 2) Dose = mg/l Dose =.5 mg/l Dose = Passive f. 2-hr Sed. With Fixed PAC Dosing (Year ) g. 2 hr Sed. With Optimized PAC Dosing (Year 2)

24 Figure Total Phosphorus Results for Non-Mechanized Filtration Systems Without Sedimentation a. Fine Sand Filter (Year ) b. Coarse Sand Filter (Year ) c. Zeolite Filter (Year ) d. Activated Alumina Filter (Year ) e. Aluminum Oxide Filter (Year ) f. Fine Sand Filter Wth Fixed PAC Dosing (Year ) g. Coarse Sand Filter With Fixed PAC Dosing (Year )

25 Figure (Part ) Total Phosphorus Results for Non-Mechanized Filtration Following Sedimentation Without Chemicals a. Concrete Sand Filter, Fast Load, Free Drain Following 2 hr Sedimentation (Year ) b. Fine Sand Filter, Fast Load, Free Drain Following 2 hr Sedimentation (Year 2) c. Fine Sand Filter, 6 hr Load, Free Drain Following 2 hr Sedimentation (Year 2) d. Fine Sand Filter, 6 hr Load, Submerged Following 2 hr Sedimentation (Year 2) e. Fine Sand Filter, 6 hr Load, Free Drain Following 24 hr Sedimentation (Year 2) f. Fine Sand Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2)

26 Figure (Part 2) Total Phosphorus Results for Non-Mechanized Filtration Following Sedimentation Without Chemicals g. Expanded Shale Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) h. Limestone Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) i. Wollastonite Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) j. Activated Alumina Filter, 6 hr Load, Submerged Following 24 hr Sedimentation (Year 2) Non-Mechanized Filtration Following Sedimentation With Chemicals Total phosphorus results for these systems, all using PAC as a coagulant, are shown in Figure 2. All of the experiments shown in this figure occurred in Year except for the fine sand with slow loading and optimized PAC dosing (Figure 2b). As can be seen, this was the only system that met the.-mg/l total phosphorus limit in all runs. Three features differentiate the fine sand system in Year 2 (Figure 2b) from that in Year (Figure 2a): optimized vs. constant dosing, slow vs. fast loading, and different storm waters. How much each factor contributed to the improvement in performance cannot be distinguished from these data. All of the other systems tested were successful in accomplishing substantial phosphorus removals. Although the results were similar for all the non-sand media tested in Year, the activated alumina and aluminum oxide filters performed slightly better than the other two, meeting the regulatory limit in four of five experimental runs. In comparing the Year fine sand filter system (2a) with the other Year systems (2c through 2f), it should be noted that the fine sand filter system was operated for six runs, whereas the other systems were operated for only five runs. The worst performance for this unit (the high point in Figure 2a) was in the sixth run when the other units were not operated. Whether this result is due to a deficiency in the sand filter, or whether it reflects some special characteristic of 375

27 the influent that would have prevented the other filters from treating it effectively, cannot be ascertained. Therefore, this data point should not be considered in comparing the systems. Figure 2 Total Phosphorus Results for Non-Mechanized Filtration Following Sedimentation With PAC a. Fine Sand Filter, Fast Load, Free Drain, Following 2 hr Sed. With Fixed PAC Dosing (Year ) b. Fine Sand Filter, 6 hr Load, Free Drain, Following 2 hr Sed. With Optimized PAC Dosing (Year 2) c. Coarse Sand Filter, Fast Load, Free Drain Following 2 hr Sed. With Fixed PAC Dosing (Year ) d. Zeolite Filter, Fast Load, Free Drain Following 2 hr Sed. With Fixed PAC Dosing (Year ) e. Activated Alumina Filter, Fast Load, Free Drain Following 2 hr Sed. With Fixed PAC Dosing (Year ) f. Aluminum Oxide Filter, Fast Load, Free Drain Following 2 hr Sed. With Fixed PAC Dosing (Year ) Mechanized Treatment Systems Total phosphorus results for the mechanized treatment systems tested in Years and 2 are shown in Figure 3. In Year 2, both the Actiflo and the conventional coagulation, flocculation, and sedimentation systems produced effluents with total phosphorus below the reporting limit in all six runs. Results for the downstream filters and ion exchange units for Year 2 are irrelevant and are not shown. In Year, the only mechanized coagulation, flocculation, and sedimentation system tested was the Actiflo system. In that year, the total phosphorus limit was not met by Actiflo treatment in 376

28 two of the five runs completed. Furthermore, for those two runs (Runs and 6), subsequent treatment by the filters and ion exchange units still failed to meet the limit (the pressure sand filter was not operated in Run 6). Only five ion exchange results are shown in Figure 3d because the ion exchange effluents for both the Fuzzy Filter and the pressure sand filter were composited together before the sample was sent out for laboratory analysis. The specific influent characteristics or other factors that prevented adequate treatment in Runs and 6 are unknown. In Run the dissolved fraction of the influent was about. mg/l and never declined as the water moved through the treatment processes. In Run 6, the influent dissolved phosphorus was an unusually high.68 mg/l, and even though it declined substantially as the water moved through the treatment processes, the drop in dissolved concentration wasn t enough to cause the total phosphorus in the effluent to drop below. mg/l. Figure 3 Total Phosphorus Results for Mechanized Treatment Systems a. Actiflo With Fixed Chemical Dosing (Year ) b. Actiflo and Fuzzy Filter With Fixed Chemical Dosing (Year) c. Actiflo and Press. Sand Filter With Fixed Chemical Dosing (Year ) d. Actiflo, Press. Sand Filter or Fuzzy Filter, and Ion Exchange With Fixed Chemical Dosing (Year ) e. Actiflo With Optimized Chemical Dosing (Year 2) f. Conventional Coagulation, Flocculation, and Sed. With Optimized Chemical Dosing (Year 2)

29 OTHER CONSTITUENTS Although turbidity and phosphorus are the most important constituents in terms of effects on Lake Tahoe, legal restrictions have been placed on nitrogen, iron, and oil and grease as well. These constituents are discussed briefly below. Also, considered in this section are impacts on dissolved aluminum and ph caused by some of the treatment systems. Total Nitrogen None of the non-mechanized or mechanized treatment systems tested in Year was successful at consistently removing total nitrogen to meet the regulatory limit of.5 mg/l. Analysis of the influent characteristics revealed that in most experimental runs, soluble organic nitrogen plus nitrite and nitrate exceeded.5 mg/l. Because none of the experimental treatment processes except ion exchange were designed to remove dissolved forms of nitrogen, their ineffectiveness is not surprising. In Year 2, several systems appeared to be successful. These results are, however, inconclusive because the influent concentrations were significantly lower than average highway runoff values. As shown in Table, highway runoff averaged 2.7 mg/l total nitrogen, while the small-scale influent concentrations never exceeded.5 mg/l and were often below mg/l. As in Year, a significant fraction of the influent nitrogen was dissolved. Details on the forms of nitrogen observed and the relative effectiveness of various systems can be found in Caltrans 23b and 23c. Total Iron The regulatory limit for total iron in storm water discharged to surface waters in the Lake Tahoe basin is.5 mg/l. Influent storm water concentrations in the pilot testing program were always substantially higher, with a median value around mg/l and a high around 33 mg/l. Despite this, all the mechanized systems and the non-mechanized systems employing chemicallyenhanced sedimentation and filtration either met or almost met the limit in the majority of experimental runs. The activated alumina and expanded shale filters following sedimentation were the only successful systems without chemical addition. Non-mechanized systems without chemicals involving sedimentation alone, filtration alone, and sedimentation followed by sand filtration (all three types of sand tested) were ineffective. Details on the performance of the various treatment systems with regard to total iron removals can be found in Caltrans 23b and 23c. Oil and Grease As indicated in Table 2, the average oil and grease concentration for Lake Tahoe storm water is 8 mg/l. In the small-scale pilot testing program, though, oil and grease were often not detected above the reporting limit in influent samples, and when it was above the reporting limit, concentrations were in the range of only 5 to mg/l. These low values may be due to partial treatment in the detention basins from which the experimental storm water was obtained and/or may be due to oil and grease adhering to the surfaces of tanks and piping during transport and storage. Generally, all of the mechanized treatment systems and most of the non-mechanized 378